Abstract: In order to improve the forecast accuracy of market price of agricultural products and find out the products with abnormal market price timely， the paper made a comparative analysis of 9 different kinds of time\|series non\|seasonal exponential forecast methods， including the weighted arithmetic average and the double exponential smoothing method based on the monthly average price of 13 kinds of products such as carrot from Lvxin agricultural products wholesale market of Jincheng city， Shanxi province， and an improved double exponential smoothing method was presented finally. The square sum of error of the improved method was no more than that of the previous ones， and the errors of the improved one decreased 3007% according to the forecast data of March. Products with the abnormal market price in this wholesale market in a certain month can be determined with an integrated analysis of a certain products square sum of error of forecasting value and actual value， as well as that of a historical forecast.

Key words: the market price of agricultural product, forecast method of double exponential smoothing, product with the abnormal market price